The present invention relates to an on-demand interface device with a window display, which executes a help function for assisting the user in operation and/or a function for performing automatic operation with a current application such as a computer application, office automatic apparatus and home electronic apparatus.
A variety of applications using the voice recognition function have been increasingly developed, most of which, however, are of the command driven type.
Japanese Laid-Open Patent Publication No. 2000-207085 discloses an on-demand interface device that displays on a screen an agent who is an impersonated character used for explaining how to operate the current application. This device receives a request text inputted by a user, retrieves suitable information from a database containing instructions or advices for operating the current application, and executes a script related to the retrieval result, causing the agent to explain the operation.
When a voice input application relies on, not a command, a dictation for inputting natural language, the voice recognition accuracy is decreased than that of command-relied system and the further processing is affected. While an enroll function provided by a voice recognition engine may improve the recognition accuracy, the concept of enrolling is not known itself to common users who hardly recognize in practice the fact that the application could not operate well due to the decreased accuracy of the voice recognition.
The application cannot be satisfactory used as far as the voice recognition accuracy is not improved. For example, whenever a user inputs a spoken request, the system may repeat the same response “I cannot understand”, lowering the user's will to use the application.
For an application allowing the user to enter a command by using both a voice input means and a keyboard, the user may start inputting a speech by using a voice input button or a keyed command by pressing a key input button. However, a dialogue cannot be always displayed on a screen. In the application, any button cannot be displayed on the screen while an agent in shape of an impersonated character is displayed thereon. One of the solutions is to assign commands “voice input start” and “keyboard input start” to keys on the keyboard. However, there is a fear of confusing users in using two similar keys.
In case of explaining how to operate the current application by the agent being always displayed on a screen, it is unclear when the explanation is finished. Therefore, it is desirable to cause the agent to appear at the beginning of explanation and disappear at the end of explanation.
With an interface using only an agent explaining the operation method without a dialogue indicating retrieved results, there may be such a problem that, when the user wants to hear again the same explanation by the agent, he/she must enter the same instruction to cause the system to repeat the same retrieval operation. This interface may draw the attention of the user to the agent and hence cause the user to pass over other indication such as a retrieve object mode, a volume level of a voice input, an edit box of a keyboard input, which may be displayed on specified positions or in separate windows on a screen irrespective of the agent.
There is still left such a problem that the user who is first use the on-demand interface may be puzzled what to do next after turning on the interface device itself.